11:00 AM - 1:00 PM
[MIS10-P06] Assessment of Pre-earthquake Information in Ionospheric Electron Density Data Observed by Ionosonde, at Kokubunji, Japan
Keywords:Assessment of Pre-earthquake Information, Ionospheric Electron Density Data, Ionosonde, Molchan’s Error Diagram analysis
Recently, ionospheric anomalies related to earthquakes have been reported and are considered promising for short-term earthquake forecast. In Japan region, the previous statistical studies on GNSS-TEC and Ionosonde show a statistically significant correlation between sizable earthquake and TEC anomaly about 0-5 days before the earthquake. Similarly, regarding the NmF2 by ionosonde, there is a significant correlation between them on 1-10 days before the earthquakes with M<6, depth D<40 km, epicenter distance R<350 km. Here, NmF2 indicates the maximum electron density in Ionosphere.
In this paper, we assess the effectiveness in short-term earthquake forecast using the NmF2 data during 1958-2019. In this study, we used ionosonde data observed at Kokubunji (35.71N, 139.49N), Japan, operated by the National Institute of Information and Communications Technology (NICT).
We defined the anomaly of NmF2 as the value in excess of 15 days backward median + 1.5 IQR of the NmF2 at the same time hour in the previous 15 days. And we define the anomalous day as ten or more hours of the anomalies appear in one day.
Then, we performed Molchan’s Error Diagram (MED) analysis to evaluate the efficiency of NmF2 anomalies for earthquake forecasting. MED analysis is very similar to ROC (receivers operation characteristic curve) approach. The MED results show that NmF2 anomalies contains a precursor information and the tendency is clearer for the shallower, the larger, and the closer earthquakes. Using MED analysis, we can estimate the optimum parameters of earthquake detection; lead time T, alarm window L, earthquake magnitude, hypocenter depth, epocentral distance from the ionosonde. The result indicated that the forecast is most efficient when M>6.4, D<20 km, R<200 km, T=10, L=1.Using these parameters, it is found that about 46% of targeted earthquakes can be successfully detected.
In this paper, we assess the effectiveness in short-term earthquake forecast using the NmF2 data during 1958-2019. In this study, we used ionosonde data observed at Kokubunji (35.71N, 139.49N), Japan, operated by the National Institute of Information and Communications Technology (NICT).
We defined the anomaly of NmF2 as the value in excess of 15 days backward median + 1.5 IQR of the NmF2 at the same time hour in the previous 15 days. And we define the anomalous day as ten or more hours of the anomalies appear in one day.
Then, we performed Molchan’s Error Diagram (MED) analysis to evaluate the efficiency of NmF2 anomalies for earthquake forecasting. MED analysis is very similar to ROC (receivers operation characteristic curve) approach. The MED results show that NmF2 anomalies contains a precursor information and the tendency is clearer for the shallower, the larger, and the closer earthquakes. Using MED analysis, we can estimate the optimum parameters of earthquake detection; lead time T, alarm window L, earthquake magnitude, hypocenter depth, epocentral distance from the ionosonde. The result indicated that the forecast is most efficient when M>6.4, D<20 km, R<200 km, T=10, L=1.Using these parameters, it is found that about 46% of targeted earthquakes can be successfully detected.